Re: [gtk-list] Re: g_malloc() aborts when enough memory is not found?

["Emmanuel DELOGET" <logout free fr>]

> Mikko Rantalainen <mira@st.jyu.fi> writes:
> [snipped - sorry]
>
> If you need non-aborting behavior, use malloc() ;-)

    Hi Owen,

    I don't know if this is a good idea. If it is possible
    to extend the callback architecture to glib (for version
    1.30, perhaps :)), the abort behavior should be avoided in
    some cases. Consider the following (in a multiple window
    text processor)

    enum
    {
        GALLOC_WAIT,        /* do not malloc, return NULL on g_malloc()
error */
        GALLOC_NOWAIT,  /* try to alloc again */
        GALLOC_ERROR      /* cannot do anything, so g_malloc() aborts */
    };

    /* during init */
    guint granularity = 64 * 1024 ; /* ask for 64ko on a malloc error */
    gallocator_init(my_callback_for_the_no_more_memory_error, granularity);

    /* lots of g_malloc and then...*/

    gint my_callback_..._error(guint requested_size)
    {
        if (number_of_opened_docs > 1)
        {
            show_simple_dialog("no more memory : you should close some
documents");
            return (GALLOC_WAIT);
        }
        else
        {
            show_simple_dialog("fatal : no more memory");
            return (GALLOC_ERROR);
        }
    }

    Does not seems to be bad for me :)

    Emmanuel

>
> g_malloc() is basically a wrapper around malloc() that aborts
> if memory cannot be allocated.
>
> [ Right now, there is also the ability to add some simple checking
>   into the g_malloc() with compile time options to g_malloc() -
>   so you should not mix malloc() and g_free() or vice-versa.
>
>   Use of these features is discouraged and they will most likely
>   be removed in the feature. ]
>
> Regards,
>                                         Owen